Vertex AI 端点 - 无法创建端点
我正在尝试使用 Vertex AI 将自定义模型部署到端点。我运行了自定义训练,并且使用 Tensorflow 2 export_saved_model
作为估计器在我的存储桶中正确创建了模型。此存储桶中有 saved_model.pb
文件以及文件夹 variables
。 但是,当我尝试创建端点选择已保存模型的路径时,会发生以下错误:
由于错误而无法创建端点“endpoint_name”:APPLICATION_ERROR; google.cloud.ml.v1/ModelService.CreateVersion;字段:version.deployment_uri 错误:部署目录 gs:// Different_bucket/artifacts/ 预计仅包含以下之一:[saved_model.pb、saved_model.pbtxt]。
看起来它正在不是我设置的存储桶中搜索 .pb 文件。 建议?
I'm trying to deploy a custom model to an endpoint with Vertex AI. I run the custom training and the model was correctly created in my bucket using Tensorflow 2 export_saved_model
for estimators. In this bucket there is the saved_model.pb
file with the folder variables
.
However, when I try to create an endpoint selecting the path to the saved model, the following error occurs:
Failed to create endpoint "endpoint_name" due to error: APPLICATION_ERROR; google.cloud.ml.v1/ModelService.CreateVersion;Field: version.deployment_uri Error: Deployment directory gs://different_bucket/artifacts/ is expected to contain exactly one of: [saved_model.pb, saved_model.pbtxt].
It seems it is searching the .pb file in a bucket that is not the one I set.
Suggestions?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
我解决了在端点创建期间更改要传递的文件夹的问题。我的模型位于以下路径中:
gs://my_bucket/my_folder/model/saved_model.pb
而不是指向
gs://my_bucket/my_folder/model/
我使用了gs://my_bucket/my_folder
并且它有效。I solved the problem changing the folder to pass during the endpoint creation. I have the model in the following path:
gs://my_bucket/my_folder/model/saved_model.pb
Instead of pointing to
gs://my_bucket/my_folder/model/
I usedgs://my_bucket/my_folder
and it worked.